Docs

Put metre-2 in front of any chat-completions client. Compression is automatic, deterministic, and cache-safe (<50ms).

from token_metre import wrap
client = wrap(OpenAI())

Questions: team@tokenmetre.store